Language Model Checkpoints & Variants

10 repos

Pre-trained and fine-tuned language model checkpoints across multiple architectures (Mamba, Pythia, ModernBERT) at various scales, typically distributed via safetensors format. This cluster represents model weight repositories and their variants rather than training frameworks or inference libraries—practical resources for practitioners deploying or adapting existing language models. The central repos span both base pretrained versions and task-specific fine-tuned variants, offering a range of model sizes from 790M to 1.4B parameters.